摘要: 现有的矩形识别方法存在检测速度慢、识别率低和精度不足等问题,不能满足现代工业在线检测发展的需求。为此,提出一种灰度投影积分极值法的矩形检测方法,把待检测图像沿0~π方向进行投影得到灰度投影积分(GPI)矩阵,求得GPI矩阵中极值点的位置,将其保存为新的矩阵GPIEV,并在GPIEV矩阵中搜索平行线重构矩形。实验结果表明,该方法具有较好的抗干扰能力,且运算速度快,能够满足实时处理的要求。
关键词:
灰度,
极值,
矩形检测,
图像处理,
Hough变换,
在线检测
Abstract: Most of the rectangle recognition methods due to the low speed of detection, low recognition rate and lack of precision, they can not meet the needs of modern industrial development. To solve the above problems, a gray projecting integral extreme value method to detect rectangular is presented. It gets the image gray value projected in all directions to form a Gray Projection Integral(GPI) matrix, obtains the GPI matrix extreme points saved as a new matrix GPIEV, and searches for parallel lines in the GPIEV matrix to reconstruction rectangle. Experimental results show that the method has better anti-jamming ability, quick operation speed, and it can meet the requirement of real-time processing.
Key words:
gray,
extreme,
rectangle detection,
image processing,
Hough transform,
online detection
中图分类号:
任玲辉, 刘凯, 张海燕, 杨慧云. 灰度投影积分极值法的矩形检测[J]. 计算机工程, 2012, 38(08): 159-160.
LIN Ling-Hui, LIU Kai, ZHANG Hai-Yan, YANG Hui-Yun. Rectangle Detection of Gray Projection Integral Extreme Value Method[J]. Computer Engineering, 2012, 38(08): 159-160.